Enhancing collaboration between humans and ChatGPT-3 through effective prompting: Revolutionizing EFL instruction in Moroccan secondary schools

Abstract

This mixed-methods study examines how Moroccan EFL teachers implement prompt engineering to design lesson plans using Large Language Models (LLMs), particularly ChatGPT, while maintaining alignment with the OHEE scheme. The research methodology combined a quantitative survey (n=249) analyzing teachers' interaction patterns with these tools, followed by qualitative focus group discussions (n=20) conducted in two phases.

The study used purposive sampling to ensure demographic variety and representation across different teaching environments. In the initial phase, teachers without training (n=10) created basic prompts, producing eight first attempts at crafting instructional queries. The second phase involved trained teachers (n=10) in a refined prompt development process, generating 42 prompts across seven conversation-based interactions with ChatGPT. The training introduced a structured framework incorporating five essential elements: persona, audience, context, specific instruction, and output specification.

Quantitative data analysis through SPSS identified usage patterns and behavioral trends, while qualitative analysis examined prompt quality and OHEE scheme alignment. The comparison of 50 prompts (8 basic, 42 refined) showed marked differences between the capabilities of trained and untrained teachers. Results showed that prompts from teachers without training often missed key elements, leading to misaligned educational materials, while trained teachers demonstrated superior ability to create focused, objective-aligned prompts through systematic improvement.

This research advances our understanding of technology-assisted instructional design by demonstrating the value of structured prompt engineering training in educational settings. The findings highlight the importance of systematic professional development in optimizing the use of these tools and point to new directions for research in computer-assisted language teaching methods.